[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH 5/5] xen: Write CR0, CR3 and CR4 in arch_set_info_guest()

> I took the suggestion to mean at the time that we should have something
> like EVENT_CR3_PRE and EVENT_CR3_POST, where basically all we needed was
> for all events for which this applicable to be pre-write events. IMHO
> that's simpler and sufficient: just send out an event when you know
> that, unless you deny it, simply responding to it / unblocking the VCPU
> will perform the write. So you know that the write is about to happen,
> and by no denying it, that it will happen (or at least, that it's
> extremely likely to happen - since some HV check can still fail somewhere).
> So, again, just IMHO, the simpler modification would just be to turn all
> events where this is applicable into pre-write events.

Isn't the event from the guest's perspective guaranteed to be
pre-write already? IMHO there is not much point in having two distinct
event types (PRE/POST). I'm not particularly happy with the "deny
change" flag idea either. Why not let the user specify what value he
wants to set the register to in such a case? It could the one that was
already there (old_value) in "deny change" style, but it might as well
be something else. I'm thinking we could keep the existing vm_event
hooks in place and simply let the vm_event response specify the value
the register should be set to (in the new_value field).


Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.